In this thesis, we established a video surveillance system that detects events
important to the user within a home network using computer vision and notifies
the user about them. A significant feature of the system is that all data is stored
and processed locally, avoiding the need to send data to the cloud and preventing
potential data misuse.
First, we set up a home server running the Home Assistant operating system.
To this operating system, we added several extensions, with the primary extension
being Frigate NVR, which is responsible for event detection. We configured the
Frigate NVR extension according to our specific requirements to achieve optimal
results.
Using the Duck DNS service, we enabled access to our server from outside
the local network, allowing the user constant access to the system. For sending
notifications, we chose the Telegram platform, as it offers easy integration with
Home Assistant and is freely accessible.
The results of the thesis show that it is possible to establish a reliable and
efficient video surveillance system using affordable components and open-source
software. The main conclusions indicate that the system is suitable for home use
and allows for easy expansion and customization according to the user’s specific
needs.
|
